Skip to content
1337skills 1337skills

Flowise Cheat Sheet

Overview

Flowise is an open-source low-code tool for building LLM applications through a visual drag-and-drop interface. Built on top of LangChain and LlamaIndex, it enables rapid prototyping of chatbots, RAG systems, and multi-agent workflows without writing extensive code. Flowise supports dozens of LLM providers, vector stores, document loaders, and tools through its modular node system.

The platform is designed for developers and non-developers alike who want to experiment with and deploy AI pipelines quickly. It provides a marketplace of pre-built templates, built-in credential management, conversation memory, and REST API endpoints for every chatflow. Flowise can be self-hosted or deployed to cloud platforms with Docker.

Installation

NPM (Quick Start)

npm install -g flowise
npx flowise start
# Access UI at http://localhost:3000

Docker

docker run -d \
  --name flowise \
  -p 3000:3000 \
  -v flowise_data:/root/.flowise \
  flowiseai/flowise

# With environment variables
docker run -d \
  --name flowise \
  -p 3000:3000 \
  -e FLOWISE_USERNAME=admin \
  -e FLOWISE_PASSWORD=securepassword \
  -e APIKEY_PATH=/root/.flowise \
  -e DATABASE_PATH=/root/.flowise \
  -v flowise_data:/root/.flowise \
  flowiseai/flowise

Docker Compose

version: '3.8'
services:
  flowise:
    image: flowiseai/flowise
    restart: always
    ports:
      - "3000:3000"
    environment:
      - FLOWISE_USERNAME=admin
      - FLOWISE_PASSWORD=securepassword
      - DATABASE_TYPE=postgres
      - DATABASE_HOST=postgres
      - DATABASE_PORT=5432
      - DATABASE_USER=flowise
      - DATABASE_PASSWORD=flowise
      - DATABASE_NAME=flowise
    volumes:
      - flowise_data:/root/.flowise
    depends_on:
      - postgres

  postgres:
    image: postgres:16
    restart: always
    environment:
      - POSTGRES_USER=flowise
      - POSTGRES_PASSWORD=flowise
      - POSTGRES_DB=flowise
    volumes:
      - pgdata:/var/lib/postgresql/data

volumes:
  flowise_data:
  pgdata:

From Source

git clone https://github.com/FlowiseAI/Flowise.git
cd Flowise
pnpm install
pnpm build
pnpm start

Core Concepts

Node Categories

CategoryExamplesPurpose
LLMsChatOpenAI, ChatAnthropic, OllamaLanguage model providers
EmbeddingsOpenAI Embeddings, HuggingFaceText vectorization
Vector StoresPinecone, Chroma, Qdrant, FAISSStore and retrieve vectors
Document LoadersPDF, CSV, Web, Notion, ConfluenceIngest documents
Text SplittersRecursive, Token, MarkdownChunk documents
MemoryBuffer, Window, Summary, RedisConversation history
ChainsConversational, RetrievalQA, LLMProcessing pipelines
AgentsOpenAI Functions, ReAct, Plan-ExecuteAutonomous tool use
ToolsCalculator, SerpAPI, Custom APIAgent capabilities

API Usage

# Send a prediction (chat message)
curl -X POST http://localhost:3000/api/v1/prediction/{chatflowId} \
  -H "Content-Type: application/json" \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "question": "What is machine learning?",
    "overrideConfig": {
      "temperature": 0.7
    }
  }'

# Send with conversation history
curl -X POST http://localhost:3000/api/v1/prediction/{chatflowId} \
  -H "Content-Type: application/json" \
  -d '{
    "question": "Tell me more about that",
    "chatId": "existing-chat-id",
    "history": [
      {"role": "user", "content": "What is ML?"},
      {"role": "assistant", "content": "ML is..."}
    ]
  }'

# Upload file for processing
curl -X POST http://localhost:3000/api/v1/prediction/{chatflowId} \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -F 'files=@document.pdf' \
  -F 'question=Summarize this document'

# Upsert documents to vector store
curl -X POST http://localhost:3000/api/v1/vector/upsert/{chatflowId} \
  -H "Content-Type: application/json" \
  -d '{
    "overrideConfig": {
      "pineconeIndex": "my-index"
    },
    "stopNodeId": "vectorStore_0"
  }'

# List chatflows
curl -X GET http://localhost:3000/api/v1/chatflows \
  -H "Authorization: Bearer YOUR_API_KEY"

Building a RAG Pipeline

Step-by-Step Flow

  1. Document Loader — Add a PDF Loader or Text File node
  2. Text Splitter — Connect a Recursive Character Text Splitter
  3. Embedding — Add OpenAI Embeddings or HuggingFace
  4. Vector Store — Connect to Pinecone, Chroma, or FAISS
  5. Retriever — Configure top-k and similarity threshold
  6. LLM — Add ChatOpenAI or ChatAnthropic
  7. Chain — Use Conversational Retrieval QA Chain

Override Configuration

{
  "question": "Explain the architecture",
  "overrideConfig": {
    "model": "gpt-4",
    "temperature": 0.3,
    "maxTokens": 2000,
    "topK": 5,
    "systemMessage": "You are a technical documentation assistant."
  }
}

Configuration

Environment Variables

VariableDescriptionDefault
FLOWISE_USERNAMELogin usernameNone
FLOWISE_PASSWORDLogin passwordNone
PORTServer port3000
DATABASE_TYPEsqlite or postgres or mysqlsqlite
DATABASE_PATHSQLite database path~/.flowise
APIKEY_PATHAPI keys storage path~/.flowise
LOG_LEVELerror, info, verbose, debuginfo
TOOL_FUNCTION_BUILTIN_DEPAllowed npm packages in custom toolsNone
TOOL_FUNCTION_EXTERNAL_DEPExternal npm packages whitelistNone
CORS_ORIGINSAllowed CORS origins*
IFRAME_ORIGINSAllowed iframe originsNone

Credential Management

# Credentials are stored encrypted in the database
# Set encryption key via environment variable
SECRETKEY_OVERWRITE=your-256-bit-secret-key

# Supported credential types:
# - OpenAI API Key
# - Anthropic API Key
# - Pinecone API Key
# - HuggingFace Access Token
# - AWS credentials
# - Google Cloud credentials
# - Custom API keys

Advanced Usage

Custom Tool Functions

// Custom JavaScript tool node
// Available: $input, $vars, $flow
const fetch = require('node-fetch');

const response = await fetch('https://api.example.com/data', {
  method: 'POST',
  headers: {
    'Content-Type': 'application/json',
    'Authorization': `Bearer ${$vars.apiKey}`
  },
  body: JSON.stringify({ query: $input })
});

const data = await response.json();
return JSON.stringify(data);

Embedding Chatbot Widget

<!-- Fullpage chatbot -->
<script type="module">
  import Chatbot from 'https://cdn.jsdelivr.net/npm/flowise-embed/dist/web.js';
  Chatbot.initFull({
    chatflowid: 'your-chatflow-id',
    apiHost: 'http://localhost:3000',
    theme: {
      button: { backgroundColor: '#3B81F6' },
      chatWindow: {
        welcomeMessage: 'Hello! How can I help?',
        backgroundColor: '#ffffff',
        fontSize: 16
      }
    }
  });
</script>

<!-- Bubble chatbot -->
<script type="module">
  import Chatbot from 'https://cdn.jsdelivr.net/npm/flowise-embed/dist/web.js';
  Chatbot.init({
    chatflowid: 'your-chatflow-id',
    apiHost: 'http://localhost:3000',
    chatflowConfig: {
      topK: 3,
      temperature: 0.5
    }
  });
</script>

Multi-Agent Setup

Flow structure:
  Supervisor Agent
    ├── Research Agent (SerpAPI tool)
    ├── Writer Agent (LLM + prompt)
    └── Code Agent (Code interpreter tool)

Each agent node:
  - Has its own system prompt
  - Can use specific tools
  - Reports back to supervisor
  - Supervisor decides next step

Streaming Responses

// Server-Sent Events streaming
const response = await fetch(
  'http://localhost:3000/api/v1/prediction/chatflow-id',
  {
    method: 'POST',
    headers: { 'Content-Type': 'application/json' },
    body: JSON.stringify({
      question: 'Explain quantum computing',
      streaming: true
    })
  }
);

const reader = response.body.getReader();
const decoder = new TextDecoder();

while (true) {
  const { done, value } = await reader.read();
  if (done) break;
  const chunk = decoder.decode(value);
  process.stdout.write(chunk);
}

Troubleshooting

IssueSolution
Port 3000 already in useSet PORT=3001 environment variable
Cannot connect to LLMCheck API key in Credentials, verify model name
Vector store upsert failsEnsure embeddings dimension matches vector store config
Memory overflow with large docsUse smaller chunk sizes (500-1000 chars)
CORS errors in embed widgetSet CORS_ORIGINS to your domain
Custom tool npm packages failAdd package names to TOOL_FUNCTION_EXTERNAL_DEP
SQLite database lockedSwitch to PostgreSQL for production
Chatflow not respondingCheck node connections, ensure all required inputs are linked
# Debug logging
LOG_LEVEL=debug npx flowise start

# Check container logs
docker logs -f flowise

# Reset database
rm -rf ~/.flowise/database.sqlite
npx flowise start